An AI-driven clinical care pathway to reduce 30-day readmission for chronic obstructive pulmonary disease (COPD)...

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Pulmonology

Background:

• Healthcare agencies mandate reduced 30-day hospital readmissions.
• Chronic Obstructive Pulmonary Disease (COPD) significantly contributes to readmission rates.
• Early identification of high-risk patients is crucial for intervention.

Purpose of the Study:

• To develop and validate an Artificial Neural Network (ANN) tool for early identification of COPD patients at high risk of 30-day readmission.
• To implement the ANN model in a smartphone application for clinical use.
• To assess the impact of the tool and subsequent interventions on COPD readmission rates.

Main Methods:

• Utilized COPD patient data from eight hospitals to identify four key predictive variables: prior admissions, first-day medications, insurance status, and Rothman Index.
• Trained an ANN model to create a predictive algorithm, validated on a separate dataset.
• Developed the Re-Admit smartphone app incorporating the ANN model and implemented targeted clinical care plans for high-risk patients.

Main Results:

• The ANN model achieved an Area Under the Curve (AUC) of 0.77, with 0.75 sensitivity and 0.67 specificity for predicting readmission.
• Implementation of the Re-Admit app and clinical interventions resulted in a significant 48% decline in readmission rates within the high-risk COPD subgroup.
• The AI-enabled app accurately predicts readmission risk on day one, enabling timely resource allocation.

Conclusions:

• The ANN model demonstrates efficacy in predicting readmission risks for COPD patients.
• The AI-enabled Re-Admit smartphone app facilitates early intervention, optimizing clinical care pathways.
• This approach effectively reduces hospital readmissions for high-risk COPD patients.